Plugs and chugs—seismic and acoustic observations of ... · canoes, it is likely that mechanisms...

Plugs and chugs—seismic and acoustic observations of degassingexplosions at Karymsky, Russia and Sangay, Ecuador

J.B. Johnsona,* , J.M. Leesb

aGeophysics Program, University of Washington, Seattle, WA 98195, USAbDepartment of Geology and Geophysics, Yale University, New Haven, CT 06510, USA

Received 30 November 1999

Abstract

Frequent degassing explosions, occurring at intervals of minutes to tens of minutes, are common at many active basaltic andandesitic volcanoes worldwide. In August 1997, April 1998, and September 1998 we recorded seismic and acoustic signalsgenerated at two andesitic volcanoes with ‘Strombolian-type’ activity. Despite variations in explosion frequency (5–15 h21 atKarymsky as opposed to 1–3 h21 at Sangay), the signatures of the explosions are remarkably similar at these two, diverse fieldsites. In all explosions, gas emission begins rapidly and is correlated with an impulsive acoustic pressure pulse. Seismicwaveforms are emergent and begin 1–2 s before the explosion. We classify explosion events at the two volcanoes as eithershort-duration (less than 1 min) simple impulses or long-duration (up to 5 min) tremor events. Many tremor events haveharmonic frequency spectra and correspond to regular 1 s acoustic pulses, often audible, that sound like chugging from alocomotive. Chugging events are intermittent, suggesting that the geometry or geochemistry of the process is variable over shorttime scales. We attribute the 1 Hz periodic chugs to a resonant phenomenon in the upper section of the conduit.q 2000 ElsevierScience B.V. All rights reserved.

Keywords: degassing explosions; Strombolian-type activity; seismo-acoustic signals

1. Introduction

The broad spectrum of volcanic behaviors world-wide can be attributed to variable magma chemistries,which govern the volatile contents and viscosities ofmagmas. Because time scales and explosivities rangeover many orders of magnitudes at different vol-canoes, it is likely that mechanisms for volcanicdegassing are highly variable. We focus on activityat andesitic volcanic centers with Strombolian-typeactivity because this type of activity is interesting,common, and poorly understood (Simkin, 1994;

Sparks, 1997). Because degassing events are frequentand relatively small (VEI 2), this type of volcanooffers ideal opportunities for seismic field analysis.In this paper we present data and observations fromthree field experiments at two volcanoes, Karymskyand Sangay, and introduce physical models to explainthe phenomena. Karymsky and Sangay both possessmiddle-of-the-spectrum volcanic characteristicsincluding: eruptive time scale (many discrete explo-sions each day), magma viscosity (103–105 Pa s),volatile content (3–5 weight percent at depth), andmass flux through vent (less than 103 kg/s) (Simkin,1994; Sparks, 1997).

In 1997 and 1998, we conducted similar seismo-acoustic experiments at two different andesitic

Journal of Volcanology and Geothermal Research 101 (2000) 67–82

0377-0273/00/$ - see front matterq 2000 Elsevier Science B.V. All rights reserved.PII: S0377-0273(00)00164-5

www.elsevier.nl/locate/jvolgeores

* Corresponding author. Fax:1 1-206-543-0489.E-mail address:[email protected] (J.B. Johnson).

volcanoes with Strombolian-type activity: KarymksyVolcano in Kamchatka, Russia and Sangay Volcanoin the Northern Andes, Ecuador. Prior to our fieldefforts, neither of these volcanoes had been thefocus of detailed seismo-acoustic studies. At both vol-canoes, activity was characterized by summit ventexplosions, which occurred several times each hour,releasing gas and ejecting small amounts of ash andballistics. Such explosions are typical of activity at thetwo volcanoes during historic times (Hall, 1977;Ivanov et al., 1991).

Degassing at Karymsky and Sangay is similar inmany ways to degassing at other volcanoes, such asArenal in Costa Rica (Benoit and McNutt, 1997),Langila in Papua New Guinea (Mori et al., 1989),and Semeru in Indonesia (Schlindwein et al., 1995).At these sites, discrete explosive events, audible‘chugs’, intermittent lava flows, and an andesitic orbasaltic–andesitic magma composition have beendocumented. We consider the regular explosions atvolcanoes such as Stromboli, Kilauea, Villarica, andErebus to be less analogous because they possessgenerally lower magma viscosities. Though regular,frequent explosions also characterize activity at thesecenters, the observations of large gas bubbles risingand bursting at the surface of lava lakes indicatesunobstructed degassing without a significantmagmatic flux (Sparks et al., 1997; Global VolcanismProgram, 1999).

2. Background

Karymsky Volcano, located in the central portionof Kamchatka’s main active arc, began erupting inJanuary 1996 after 14 years of quiescence (Gordeevet al., 1997). Though vigorous Vulcanian activitymarked the eruption onset, activity quickly declinedto discrete explosive events. In the summer of 1996,about 12 explosions occurred each hour, while only 5per hour occurred in the summer of 1997. A blocklava flow, which was energetic in 1996, was barelymoving in 1997. In 1998, the lava flow remobilizedand explosions were again more frequent. Though theratio of ejected magma to gas is variable, each explo-sion appears to contain a ballistic or ash component,indicating that degassing may not be possible withouta magmatic flux. Karymksy is a young (4 ka), symme-

trical 800-meter-high cone which rises above the floorof an older caldera (Ivanov et al., 1991). A singlesummit vent is the source of both the lava flows andexplosive degassing. Typical composition of recentKarymsky lava flows is 62.20 weight percent silica(Ivanov et al., 1991).

Sangay Volcano, located in Ecuador’s easternvolcanic cordillera, has been continuously activesince 1628 (Hall, 1977). However, due to its remoteposition, fluctuations in eruptive vigor may go largelyunnoticed. During the spring of 1998, volcanic activ-ity was at a relative ebb, with explosive gas releasesoccurring only twice an hour. According to localobservers, typical activity is manifested by morepowerful explosions, including significant incandes-cant ejecta and extrusion of lava. During 1998, vol-canic bombs and some summit incandescenceindicated a continued, though slight, flux of magmaticmaterial through the vent. Although the compositionof Sangay lavas are widely variable, they commonlylie between 55 and 57 wt% silica (Monzier et al.,1999). The 1800-meter-high Sangay edifice is a14 ka accumulation of lava flows, debris flows, andpyroclastic flows, atop remnants of older, collapsedvolcanic structures (Monzier et al., 1999). At present,four craters are aligned along a 700 m summit ridge.During the 1998 field work, fumaroles were active inall craters, but a single central crater was the sourcefor all explosions.

In all experiments at Karymsky and Sangay, werecorded the regular explosions with Guralp CMG-40T broadband seismometers as well as bandpassed(2 s to 30 Hz) electret condenser microphones. Atboth volcanoes, microphones and seismometerswere deployed within 1.5 km of the active vent. Ateach volcano, digital recording was continuous for atleast 100 h. Seismic signal-to-noise is excellent, whileacoustic quality is variable and dependent upon lowwind conditions. At both volcanoes, we were able torecord at least 24 h of high signal-to-noise acousticexplosion signals.

3. Data overview

We conducted field work at Karymsky duringAugust 1997 (Johnson et al., 1998), and September1998. In 1997, we deployed three 3-component broad-

J.B. Johnson, J.M. Lees / Journal of Volcanology and Geothermal Research 101 (2000) 67–8268

J.B. Johnson, J.M. Lees / Journal of Volcanology and Geothermal Research 101 (2000) 67–82 69

Fig. 1. (a) Map of Karymksy station installations, 1997. Station KAR1 operated for 9 days and contained a broadband seismometer and low-frequency microphone. Stations Kar2 through Kar7 contained broadband seismometers and recorded for two days each. Inset map showslocation of Karymsky Volcano within the Kamchatka Peninsula. (b) Map of Sangay and station installations in 1998. Stations San1 and San2(large triangles) contained broadband seismometers and low-frequency microphones. Small triangles indicate locations of single-componentshort period sensors. Inset map shows location of Sangay within South America. (c) Map of Karymsky and station installations in 1998. StationKry1 operated for 6 days and was the site of the Larsen–Davis precision infrasonic pressure sensor. Stations Kry2, Kry2a, and Kry3 eachcontained broadband seismometers and low-frequency microphones. StationsR1, R2, R3, R4, L2, L3, L4, V1, V2, V3 and V4 operated in one-day or two-day long campaigns and consisted of short-period instruments and low-frequency microphones.

band seismometers on the flanks of the cone (Fig. 1a).Station Kar1, located 600 m below and 1500 m southof the vent, consisted of a seismometer and alow-frequency acoustic microphone. Two other

broadband stations were moved about the volcanoevery two days in order to record seismic energypropagating through various sections of the cone. In9 days, we recorded about 1000 explosions, and more


Fig. 1. (continued)


Fig. 1. (continued)

than 100 high-quality acoustic signals. In 1998, wereturned to Karymsky with three broadbandinstruments, four short period instruments, sevenlow-frequency microphones, and one Larsen–Davis

free field precision microphone. Station Kry1 reoccu-pied the site of Kar1 and was active for the duration ofthe experiment. Six other seismo-acoustic stationswere deployed in variable geometries in day-long


Fig. 2. (a) Vertical 8-hour velocity seismogram for station Kar1 at Karymksy (21 August 1997). (b) Vertical 8-hour velocity seismogram forstation San1 at Sangay (25–26 April 1998). (c) Vertical 8-hour velocity seismogram for station Kry1 at Karymsky (6 September 1998).


Fig. 3. Spectrograms of seismic and acoustic traces for typical events at: (a) Karymsky in 1997 (recorded at Kar1); (b) Sangay in 1998 (recordedat San1); and (c) Karymsky in 1998 (recorded at Kry1). Harmonic tremor is evident in both seismic and acoustic channels. It is present in someevents from Karymsky 97, common in the Sangay 98 record, and absent at Karymsky in 98.


Fig. 4. Seismic traces and spectrograms for three characteristic event types at Karymsky in 1997: (a) high-frequency event; (b) chugging(harmonic tremor) event; (c) simple impulse event. The spectral evolution plot to the left demonstrates how specific event types tend to cluster intime for the Karymsky 97 record. Spectrograms were calculated for 80 consecutive events and used time windows 30–90 s after the explosiononset.

campaigns (Fig. 1c). Due to elevated eruptive vigor,we recorded about 2000 explosions and 1000 high-quality acoustic signals in 6 days.

We conducted field work at Sangay during April1998. Two stations (San1 and San2), containingbroadband seismometers and low-frequency micro-phones, were situated on the southern flank of thevolcano, 1200 and 1600 m from the vent (Fig. 1c).In addition, three short period vertical Mark ProductsL22 sensors were deployed linearly at 300 m inter-vals. In total, we recorded about 300 explosions,more than 50 of them associated with high-qualityacoustic records.

During our field work at Karymsky and Sangay,

eruptive activity consisted of distinct, frequent explo-sions with largely aseismic recharge intervals. Thelack of background tremor contrasts with observationsat many other volcanoes with Strombolian-type activ-ity (McNutt, 1986; Hagerty et al., 1997; Dawson et al.,1998). Our seismic records also show no evidence ofvery long period seismicity associated with masstransfer (Neuberg et al., 1994; Chouet and Dawson,1997).

Seismograms (Fig. 2a–c) and acoustograms revealsignificant similarities and some differences in erup-tive behavior among the three datasets. Virtuallyevery discrete seismic event can be associatedwith an acoustic counterpart (Fig. 3a–c), indicating


Fig. 5. Similarity of chugging (harmonic tremor) events at: (a) Karymsky in 1997 (recorded at Kar1); and (b) Sangay in 1998 (recorded at San1).Acoustic harmonic tremor is manifested as a regular series of impulsive gas bursts. The presence of more integer overtones in the Sangayharmonic tremor can be explained by a higher corner frequency for composite simple impulse events.

explosive degassing at the primary vent. Bothvolcanoes possess a variety of explosion typesincluding short-duration seismic wavelets (simpleimpulses) which taper to background in less than1 min, and longer-duration ground motions (tremor)which last as long as 5 min.

4. Analysis

At Karymsky and Sangay, we define distinct explo-sion events as short-term seismic energy (10 s meansquared velocity) which is greater than the daily meanenergy and separated from a previous event by at least30 s. According to these criteria, we document 149explosions in 27 h at Karymsky in 1997 (5.5 perhour, average event length 31 s), 85 explosions in49 h at Sangay in 1998 (1.7 per hour, average eventlength 91 s), and 200 events in 15 h at Karymsky in1998 (13 per hour, average event length 25 s). Explo-sion frequency was greatest at Karymsky in 1998,least at Sangay in 1998, and appears to be inverselyproportional to the length of the degassing events. Weattribute heightened explosion frequency and shor-tened event duration (most pronounced in theKarymsky 1998 dataset) to heightened magma andgas flux (see Section 5).

The spectral and temporal character of explosionsignals reveals several broad categories of eventtypes at Karymsky and Sangay. We refer to the firstcategory (Fig. 4c) as a simple impulse, or short-dura-tion explosion, because of the absence of sustainedground motion. This event type represents the shortestpossible seismic signal associated with visible degas-sing at the vent. The associated acoustic signalconsists primarily of a short, simple 2.5 Hz impulsivepressure pulse. The second category is comprised ofextended degassing events. These events begin assimple impulse events, with an accompanying impul-sive acoustic pulse, but ground motions and acousticsignals are longlasting. Many of the extended degas-sing events contain well-defined integer overtones(Figs. 4b, 5a and b). We refer to these harmonictremor signals as ‘chugging events’ because thetremor is accompanied by audible exhalations whichsound like a locomotive. Another type of extendeddegassing signal contains significant energy above6 Hz (Fig. 4a). Visual observations indicate that this


Fig. 6. Qualitative event classifications for: (a) Karymsky 97; (b)Sangay 98; and (c) Karymsky 98. The quantity of integer overtoneswas assessed through visual inspection.

high-frequency component results from energetic‘jetting’ of material through the vent. Though manyextended degassing events are easy to classify aseither harmonic or high-frequency hybrid combin-ations are also common. Spectrograms of consecutiveexplosions at a particular volcano (Fig. 4) reveal atendency for specific event types to cluster intime.

A qualitative map of event types (Fig. 6) revealsgeneral differences in the three datasets. In general,Karymsky possesses proportionately fewer tremorevents than Sangay. Tremor at Karymsky tends tobe of shorter duration with less-developed integerovertones than at Sangay. The relative quantity ofinteger overtones in the Karymsky 97 and Sangay98 datasets can be attributed either to the stability ofthe resonance phenomena or to the propagation filterwithin the two volcanoes. Because seismic simpleimpulses are consistently more broad-band at Sangaythan at Karymsky, we should expect a greater quantityof integer overtones in Sangay tremor (Fig. 5).Extreme examples of integer overtones in tremorsignals, such as the 30 observed at Lascar Volcanoin Chile (Hellweg, 1997a), may be explained by the

convolution of a very broadband seismic impulse witha comb function. However, the absence of harmonictremor events at Karymksy in 1998 must be attributedto a lack of a resonance mechanism (this will bediscussed later).

Unless obscured by wind noise, all seismic explo-sion signals are accompanied by an infrasonic pres-sure pulse which follows the seismic signal by 3–7 sat vent distances of 1.2–3 km. Though the onset ofeach acoustic signal is impulsive, the correspondingseismic signal is very emergent. The emergent signalsmay be partially explained by weak transmissionbetween a low-velocity fluid-filled conduit and thewall rock (Dibble, 1994), or alternatively by a diffuseor emergent source within the conduit.

Indeed, we observe evidence for source motionsprior to the actual explosion at Karymsky. A tightarray (85 m spacing) at Karymsky in 1998 (Fig. 7),reveals apparent velocities of 1645 m/s for seismicfirst arrivals and 336̂ 4 m=s for acoustic arrivals.However, the travel time separation between thesetwo phases is too lengthy to be explained by a concur-rent seismo-acoustic impulsive energy source at thevent. In many explosions, we are able to observe up to


Fig. 7. A selected Karymsky 98 explosion recorded by a tight (85 m spacing) linear array. The seismic channel has been bandpassed between 2 sand 2 Hz. The first detectable seismic energy (dashed line) crosses the array at an apparent velocity of 1645 m/s. Acoustic arrivals cross thearray at 336 m/s. A concurrent seismo-acoustic impulsive source at the vent that might be responsible for the acoustic arrival would produceseismic energy that would cross the array (dash-dot line) later than the observed seismic first arrival.

2 s of pre-explosion seismicity, which may reflectfracturing or movement of material in the conduitjust prior to gas release. These motions are slightcompared with motions which occur several secondslater. The bulk of the seismic energy is generated bythe response of the volcano to the thrust of materialejected through the vent (Brodsky et al., 1999).

Even simple explosion events generate seismic

signals, which are tens of seconds long. This codashould be attributed to interactions of body andsurface waves caused by path effects (reflections,scattering, near-field effects) within the structurallycomplicated volcanic edifice (Chouet et al., 1998).Despite complicated propagation filters, we observea high degree of self-similarity for the first ten secondsof most seismic waveforms, indicating that a similarexplosion mechanism is responsible for each event(Fig. 8). Waveform correlation is much higher fortwo different events than for a single explosionrecorded at two stations only 85 m apart. By recipro-city, we infer that source location within the conduitvaries by considerably less than 85 m vertically.

Though seismic waveforms may appear similar formany different explosions, the ratio of seismic-to-acoustic energy generated by different explosions isextremely variable (Fig. 9a and c). Weather, includingwind and a temperature-dependent velocity structure,has considerable impact on the atmospheric propa-gation of acoustic energy. On the basis of raypathsand traveltimes for infrasonic energy propagating ina stratified atmosphere (Garces et al., 1998), we candetermine bounds for amplitude enhancement andattenuation upwind and downwind (Fig. 9d). AtKarymsky in 1998 we deployed microphones north,south, east and west of the vent (Fig. 9f) with theintent of filtering out the effects of weather from theacoustic records. Even after removing conceivableimpacts of weather, we still observe considerablescatter in the plot of seismo-acoustic amplitudes(Fig. 9e). Rapid changes in seismo-acoustic ampli-tudes that occur during the course of a single explo-sion event (Fig. 9b) are further evidence that changingweather is not the sole cause of variable seismo-acoustic amplitudes.

We acknowledge several mechanisms, besidesweather (Fig. 10a), which may be responsible for vari-able seismo-acoustic ratios. The depth of the explo-sion source within the conduit could affect the amount


Fig. 8. Repetitive explosion source locations and mechanisms areevidenced by the self-similarity of explosion waveforms recorded ata single station.

Fig. 9. (a) Scatter in the normalized seismic and acoustic amplitudes for a suite of 864 distinct explosion events from Karymsky 98 (recorded atKry1 and Kry2). (b) A selected Karymsky explosion event with dramatic changes in seismo-acoustic amplitude ratios (notice double pulse). (c)Scatter in the normalized seismic and acoustic amplitudes for a suite of 108 explosions (recorded at Kry1). (d) Potential variability in acousticamplitudes for 108 explosions recorded at Kry1. Error bars are determined through analysis of acoustic records from stations R1, R3, and R4. (e)Scatter in the normalized seismic and ‘corrected’ acoustic amplitudes for a suite of 108 explosions. Acoustic amplitudes are forced to remainwithin the error bars (d), but are migrated to obtain a best linear fit. Considerable scatter remains and must be attributed to factors other thanweather. (f) Map of microphone stations used for assessing the effects of weather at Karymsky.

of energy able to propagate into the atmosphere (Fig.10b). Similarly, debris, atop the explosion source,would muffle the acoustic signal by absorbing explo-sion energy as mechanical energy (Fig. 10c) (Mori etal., 1989). Finally, preferential coupling of seismicenergy into the air and into the ground might dependon the impedance of the magma (Fig. 10d) (Hagerty etal., 1997). This last method requires that a seismo-acoustic source be located within the magma anddoes not seem appropriate at Karymsky and Sangay,where we associate acoustic and seismic signals withgas expansion at the vent.

5. Discussion

Inhibited gas flux during the intervals betweendegassing explosions suggests that either gas flow isunsteady through the upper conduit or a plug temporarilyprevents continuous gas escape. Unsteady gas flowhas been considered as a mechanism for regularexplosions at many Strombolian systems (Kilauea,Stromboli) where vents appear entirely open (Jaupartand Vergniolle, 1989; Vergniolle and Jaupart, 1990;Manga, 1996). However, because the vents atKarymsky and Sangay appear to be choked withrubble and rising viscous lava, we consider the possi-bility that the upper section of conduit is depleted ofexsolved volatiles and acts as a ‘plug’. Though soli-dified lava is likely to possess open fractures throughwhich gas can continually escape, we believe that asection of conduit in the shallow subsurface couldtemporarily serve as a barrier to rising gas (Johnsonet al., 1998). When gas pressure reaches a criticalthreshold determined by the weight and cohesion ofthe loading material, an explosion breaches the plug.Following the initial outburst, a lowered lithostaticpressure promotes further gas exsolution. Detritusthat does not clear the vent can settle into the conduitatop volatile-depleted magma, setting the foundationfor the next plug.

In many explosions at Karymsky in 1997 and atSangay in 1998, degassing is not a single isolatedburst from the vent, but a sequence of regular, impul-sive gas releases, which is manifested as seismic andacoustic tremor. We identify four models that are ableto generate periodic signals and have been offered asexplanations of harmonic tremor at Strombolian-type

volcanoes. The first model, used to explain ‘chugging’oscillations at Arenal (Garces and McNutt, 1997)and Semeru (Schlindwein et al., 1995) is thepresence of a fluid body in the upper portion ofthe volcano edifice. In this model, excitation causesthe chamber to resonate, much as the standing wavesin a vibrating organ pipe. A second model involves thegeneration of von Karman vortices on the lee side ofan obstruction in a conduit. Vortex shedding has aspecific periodicity that can be predicted by the prop-erties of the fluids and dynamics of the flow (Hellweg,1997b). A third mechanism capable of generatingregular oscillations is conduit-wall resonanceproduced by non-linear fluid flow through a pipe(Julian, 1994). A final model, which is physicallysimilar to conduit-wall resonance, involves a pres-surized system of mixed phase gases and magmaticfluids trapped beneath a plug. The plug acts like avalve on a pressure cooker, allowing the release ofgas as a series of regular impulsive events (Lees andBolton, 1998).

Several mechanisms are theoretically capable ofproducing regular source pulses, which can propagatethrough the atmosphere and ground as harmonictremor. At this juncture, we admit that we cannotidentify a unique source for the harmonic tremor atour volcanoes. However, we currently favor gasrelease through a near-surface valve (plug resonance)because of the following reasons: (1) we regularlyobserve material choking the volcanic vent atKarymsky; (2) tremor is a transitory feature whichmay be governed by a continuously evolving plug;(3) tremor events appear to cluster in time; (4) tremorwas virtually absent at Karymsky in 1998, whenmagma flux was elevated and a cold viscous debrisplugs had little time to develop.

We believe that a pressure-cooker analog offers areasonable mechanism for both trapping gas betweenexplosions and producing occasional tremor events.Even during a vigorous explosion, when an entireplug might be ejected from the conduit, calving ofthe conduit walls and crater floor allows material tofall back into the conduit. When the conduit remainsunobstructed by debris, gas may escape easily as ahigh frequency jetting. Lower-frequency and harmo-nic tremor events are produced when debris is notcompletely ejected through the vent during the initialimpulse. Material is able to settle in the upper portion


of the conduit and inhibit continuous gas release. Therelatively rare incidence of tremor at Karymsky(compared with Sangay) may be attributed to amore vigorous material flux through the vent, whichkeeps the conduit open. However, other variables,such as magma viscosity, volatile content, and conduitgeometries, may also play a significant roll.

6. Conclusion

This largely observational paper describes seismicand acoustic signals associated with explosions atKarymsky and Sangay Volcanoes. In the spectrumof possible eruptive mechanisms, those of degassingat Karymsky and Sangay appear to be very similar. Atboth sites, seismo-acoustic sources are located nearthe vent and degassing occurs frequently and impul-sively. Volcanic material and debris choking theupper conduit are responsible for temporarily block-ing gas flow through the vent. The source of theharmonic tremor, which is nearly identical at thetwo volcanoes, may result from gas release througha plug, which acts as a valve. We consider the greaterquantity of tremor events at Sangay (compared toKarymsky) to be related to a lower gas and magmaflux.

Acknowledgements

We thank both the PASSCAL Instrument Centerand Brian Venema and Jim Ramey of the Universityof Washington for help with seismic and acousticinstrumentation. We also thank the Russian Academyof Sciences (Evgenii Gordeev and Alexi Ozerov) andthe Instituto Politecnica (Mario Ruiz and Pete Hall)for their helpful comments and invaluable assistancein the field. Support for this work is made possiblethrough grants from the PNSN and Yale University.

References

Benoit, J.P., McNutt, S.R., 1997. New constraints on sourceprocesses of volcanic tremor at Arenal Volcano, Costa Rica,using broadband seismic data. Geophys. Res. Lett. 24, 449–452.

Brodsky, E.E., Kanamori, H., Sturtevant, B., 1999. A seismicallyconstrained mass discharge rate for the initiation of the May 18,


Fig. 10. Cartoon of four possible mechanisms which could producevariable seismo-acoustic ratios: (a) weather variability; (b) locationof explosion source within a conduit; (c) dissipation of acousticenergy as mechanical energy; (d) variable melt impedance.

1980 Mount St. Helens eruption. J. Geophys. Res. 104, 29387–29400.

Chouet, B., Dawson, P., 1997. Observations of very-long-periodimpulsive signals accompanying summit inflation at KilaueaVolcano, Hawaii, in February 1997. Eos, Trans. Am. Geophys.Union 78, 429–430.

Chouet, B., Dawson, P., DeLuca, G., Martini, M., Milana, G.,Saccorotti, G., Scarpa, R., 1998. Array Analyses of SeismicWavefields Radiated by Eruptive Activity at StromboliVolcano. Felici, Pisa, p. 158.

Dawson, P.B., Dietel, C., Chouet, B.A., Honma, K., Ohminato, T.,Okubo, P., 1998. A Digitally Telemetered Broadband SeismicNetwork at Kilauea Volcano, Hawaii. USGS Open-File Report98-108.

Dibble, R.R., 1994. Velocity modeling in the erupting magmacolumn of Mount Erebus, Antarctica, Volcanological and Envir-onmental Studies of Mount Erebus, Antarctica. AntarcticResearch Series 66, 17–33.

Garces, M.A., McNutt, S.R., 1997. Theory of the airborne soundfield generated in a resonant magma conduit. J. Volcanol.Geotherm. Res. 79, 155–178.

Garces, M.A., Hansen, R.A., Lindquist, K.G., 1998. Traveltimes forinfrasonic waves propagating in a stratified atmosphere.Geophys. J. Int. 135, 255–263.

Global Volcanism Program, April 30, 1999. Volcanoes of theWorld. http://www.volcano.si.edu/gvp/.

Gordeev, E.I., Kasahara, M., Levina, V.I., Miyamachi, H.,Chebrov, V.N., 1997. Magma activity at Karymsky Volcanoand Academy Nauk Caldera (Kamchatka, Russia) triggers largetectonic (M7.0) event. Eos Trans. Am. Geophys. Union 78, 442.

Hagerty, M., Schwartz, S.Y., Protti, M., Garces, M., Dixon, T.,1997. Observations at Costa Rican volcano offer clues to causesof eruptions. Eos Trans. Am. Geophys. Union 78, 565.

Hall, M.L., 1977. El volcanismo en el Ecuador. Instituto Panamer-icano de Geografia e Historia, Quito, 120 pp.

Hellweg, M., 1997. Physical models for the source of harmonictremor. Eos Trans. Am. Geophys. Union 78, 430.

Hellweg, M., 1997. Listening carefully: Unique observations of

harmonic tremor at Lascar volcano, Chile. Seismol. Res. Lett.68, 319.

Ivanov, B.V., Braitseva, O.A., Zubin, M.I., 1991. Active Volcanoesof Kamchatka. Nauka, Moscow, pp. 180–203.

Jaupart, C., Vergniolle, S., 1989. The generation and collapse of afoam layer at the roof of a basaltic magma chamber. J. FluidMech. 203, 347–380.

Johnson, J.B., Lees, J.M., Gordeev, E.I., 1998. Degassing explo-sions at Karymsky Volcano, Kamchatka, Russia. Geophys.Res. Lett. 25, 3999–4042.

Julian, B.R., 1994. Volcanic tremor; nonlinear excitation by fluidflow. J. Geophys. Res. 99, 11 859–11 878.

Lees, J.M., Bolton, E.W., 1998. Pressure cookers as volcano analo-gues. Eos Trans. Am. Geophys. Union 79, 620.

Manga, M., 1996. Waves of bubbles in basaltic magmas and lavas.J. Geophys. Res. 101, 17 457–17 465.

McNutt, S.R., 1986. Observations and analysis of b-type earth-quakes, explosions, and volcanic tremor at Pavlof Vocano,Alaska. Bull. Seismol. Soc. Am. 76, 153–175.

Monzier, M., Robin, C., Samaniego, P., Hall, M.L., Mothes, P.,Arnaud, N., Cotten, J., 1999. Sangay volcano (Ecuador): struc-tural development and petrology. J. Volcanol. Geotherm. Res.90, 49–79.

Mori, J., Patia, H., McKee, C., Itikarai, I., Lowenstein, P., de SaintOurs, P., Talai, B., 1989. Seismicity associated with eruptiveactivity at Langila volcano, Papua New Guinea. J. Volcanol.Geotherm. Res. 38, 243–255.

Neuberg, J., Luckett, R., Ripepe, M., Braun, T., 1994. Highlightsfrom a seismic broadband array on Stromboli volcano. Geophys.Res. Lett. 1, 749–752.

Schlindwein, V., Wassermann, J., Scherbaum, F., 1995. SpectralAnalysis of harmonic tremor signals at Mt. Semeru volcano,Indonesia. Geophys. Res. Lett. 22, 1685–1688.

Simkin, T., 1994. Volcanoes of the World, 2nd ed. GeosciencePress, Tucson, pp. 100–111.

Sparks, R.S.J., 1997. Volcanic Plumes. Wiley, New York, 574 pp.Vergniolle, S., Jaupart, C., 1990. Dynamics of degassing at Kilauea

Volcano, Hawaii. J. Geophys. Res. 95, 2793–2809.


Plugs and chugs—seismic and acoustic observations of ... · canoes, it is likely that mechanisms...

Documents

Transcript of Plugs and chugs—seismic and acoustic observations of ... · canoes, it is likely that mechanisms...